Content uploaded by Andrew Sean Murray
Author content
All content in this area was uploaded by Andrew Sean Murray on Feb 04, 2015
Content may be subject to copyright.
GEOLOGOS 5 (2000)
PL ISSN 1426-8981 • ISBN 83-86682-45-0
Luminescence dating of sediments
using individual mineral grains
GEOFFREY A. T. DULLER1 & ANDREW S. MURRAY2
1 Institute of Geography and Earth Sciences, University of Wales,
Aberystwyth, UK
2 Nordic Laboratory for Luminescence Dating, Department of Earth
Sciences, University of Aarhus, Risø National Laboratory, DK-4000
Roskilde, Denmark
Abstract: Luminescence is widely used to produce absolute ages for the time of
deposition of a variety of types of sediments. The method relies upon the as-
sumption that all grains are exposed to sufficient daylight prior to deposition for
their luminescence signal to be reduced to a negligible level. Recent research has
focused on the analysis of the luminescence signal from single mineral grains to
produce an age. At this scale it is possible to identify different populations of
mineral grains within a sample – some of which were bleached at deposition and
some which were not. The methods involved in such analyses are discussed, and
examples are given of depositional environments where this type of analysis is
essential.
Key words: luminescence dating, laboratory technique, measurement protocols,
single grains, OSL.
Introduction
Luminescence dating is now an important tool for obtaining absolute
age estimates for sedimentary deposits. Recent reviews (e.g. Duller 1996; Aitken
1998) have highlighted the increasing use of optically stimulated luminescence
(OSL) measurements instead of thermoluminescence (TL). OSL is much better
suited to dating geological materials since the same process of optical resetting
of the luminescence signal is used in the laboratory and in nature, and less
exposure to daylight is required to reset the OSL signal than TL. However,
another major advance in the last ten years has been the adoption of new meas-
urement procedures that have permitted analysis of samples using smaller amo-
unts of material.
Luminescence dating is based on the measurement of two quantities, the
radiation dose received by a sample since burial (the palaeodose, P) and the rate
at which it has absorbed energy from the natural environment (the dose rate). Of
these, it is the palaeodose that is derived from luminescence measurements.
Dividing one quantity by the other gives the period of time since deposition of
the sediment:
Age(years) = Palaeodose(Gy)
DoseRate(Gy/year) .
An essential assumption of the method is that the luminescence signal from
a grain can be reset, or zeroed, by exposure to daylight. This is a process com-
monly termed ‘bleaching’ and occurs in many processes of erosion, transporta-
tion or deposition, particularly those that occur sub-aerially. Prior to 1991, almost
all palaeodose measurements were made using many tens of thousands of grains
of a sample, spread between many tens of sub-samples, or aliquots. An implicit
assumption within all of these methods, known as multiple aliquot methods, was
that the luminescence characteristics of each aliquot were identical. This required
either a completely homogeneous sample, or the use of aliquots of a sufficient
size to average out any variations. Where there were significant differences
between the grains, possibly due to differing extents of bleaching at deposition,
this was seen as scatter in the data and resulted in uncertainty in the final age
estimate produced (Huntley & Berger 1985). More seriously, the presence of
a few very bright grains which had not been bleached at deposition could lead to
a significant overestimate of the age. For instance, Duller et al. (1995) showed
that a glacio-fluvial deposit from Scotland yielded an OSL age that was at least
five times older than independent age estimates. Within the last 10 years methods
have been developed that can measure the palaeodose from single aliquots and,
more recently, single sand-sized mineral grains. This has important implications
for the potential of luminescence dating and its reliability. This paper outlines the
procedures involved in such measurements, and their advantages.
Single aliquot methods
A number of luminescence workers have suggested that it would be
feasible to make all the measurements necessary to calculate a palaeodose on
a single aliquot. Duller (1991, 1994a) was the first to demonstrate how this could
be undertaken in practice, and these methods have been used widely. This was
an important development for a number of reasons. Firstly, for dating geological
sediments it meant that if a sample contains a mixture of grains, some of which
had been bleached at deposition and some of which had not, this could be
identified. In such a situation, different aliquots would contain different propor-
tions of well bleached and unbleached grains and hence give different palaeodo-
G. A. T. DULLER & A. S. MURRAY
88
Fig. 1. Palaeodose plotted against the intensity of the natural signal for a set
of aliquots from (a) a well bleached sample (Bythe Lower Gravel – BLG) and
(b) a poorly bleached sample (Bythe Windermere Interstadial – BWI). In each
case the data were obtained using an additive dose single aliquot procedure
applied to potassium rich feldspars separated from late Quaternary fluvioglacial
sands from Scotland. Further details of the two sites are given in Duller (1994b).
Luminescence dating of sediments using individual mineral grains
89
ses (e.g. Li 1994). Only those grains whose luminescence signal was fully reset
at deposition would give an accurate palaeodose, and hence age; if a sample
contained some fraction of unbleached grains the palaeodose would be overesti-
mated, and so would the age. A number of different approaches have been
developed to analyse such data, both to identify those samples where this is
a problem, and in an attempt to provide an upper limit on the age estimate (e.g. Li
1994). The most important method has been to produce a scatter plot of the
values of the palaeodose and the OSL signal intensity from a number of aliquots
of the same sample. If a sample contains only well bleached grains then the
palaeodose derived from each aliquot will be similar, and there should be little
variation in signal intensity (Fig. 1a). In contrast, where a mixture of bleached
and unbleached grains are present, a wider range of palaeodoses will be obser-
ved, and those aliquots with the larger palaeodose will also tend to have a brigh-
ter luminescence signal (Fig. 1b). A linear regression line is placed through the
data, and if the slope of the line is significantly greater than zero then it is
deduced that the sample is incompletely bleached. This approach has been ap-
plied to fluvial, colluvial (Wintle et al. 1993) and glacial (Duller 1994b) sedi-
ments, but is equally applicable in other depositional environments where the
degree of optical bleaching at deposition is uncertain (e.g. coastal storm and
tsunami deposits, mass movements, soil forming processes). This form of analy-
sis relies upon an implicit assumption that within a sample all aliquots have
a similar sensitivity to radiation, and produce a fixed OSL signal per unit dose.
This assumption is reasonably valid for potassium-rich feldspars when analysing
many grains on an aliquot, and all the examples listed above conform to this
condition. However, for other materials such as quartz, or when the number of
grains in a sub-sample is small, this assumption is not valid and different me-
thods of analysis are required.
Results from small numbers of grains
Initial single aliquot work concentrated solely on the use of potas-
sium-rich feldspars as the dosimeter. However, Murray et al. (1995, 1997) and
Murray (1996) developed single aliquot analysis procedures that could be used
with quartz, a more ubiquitous mineral component of detrital sediments. Altho-
ugh the methods involved are different, the advantages of such measurements are
the same.
The most robust of the methods developed for quartz is known as the single
aliquot regenerative (SAR) dose procedure and has been described recently in
Murray & Wintle (2000). A brief summary is given here since many of the
results described in this paper have been obtained with this method. In essence
the procedure is very simple. In order to measure the palaeodose from a sample,
the natural OSL signal from an aliquot is measured. This measurement empties
G. A. T. DULLER & A. S. MURRAY
90
Fig. 2. The Single Aliquot Regenerative dose method (SAR). A series of OSL measurements are made of the natural signal (LN), and the
response to various regeneration doses (L1, L2 etc.). After each measurement, the sensitivity of the sample is measured by giving a standard
radiation dose (the ‘test dose’) and measuring the OSL signal produced (TN, T1, T2 etc.). The response of the sample to dose can be plotted on
a graph of the ratio of Lx/Tx as a function of laboratory dose. The palaeodose of the sample is calculated by interpolating the value of LN/TN
onto this response curve. For this sample, the palaeodose is 22 Gy.
Luminescence dating of sediments using individual mineral grains
91
Fig. 3. The distribution of apparent dose in young fluvial sediments from south eastern
Australia. Measurements were made on single aliquots of quartz (from Murray et al.
1995). The samples were (a) a sub-aqueous fan in a small farm dam, (b) an inchannel
flood deposit, (c) a bed channel deposit and (d) an overbank deposit.
Fig. 4. In a mixture of bleached and unbleached grains, the probability of selecting only
well-bleached grains decreases as the number of grains on an aliquot (n) increases, and as
the fraction of the grains which are unbleached increases (from Olley et al. 1999). For
many samples the proportion of grains that contribute significantly to the total luminescence
signal is small, and hence the effective value of n may be much smaller than the actual
value (see Fig. 8).
G. A. T. DULLER & A. S. MURRAY
92
the majority of the OSL from the sample. A laboratory dose can then be admini-
stered to the sample and the OSL signal (regenerated by that dose) measured.
This can be repeated a number of times with different regeneration doses in order
to characterise the way that the OSL signal grows with radiation dose. From this
response, the laboratory dose required to match the OSL signal obtained from the
natural measurement can be calculated (Fig. 2). This is called the equivalent dose
(DE) or the palaeodose (P).
In practice, there are additional complications. The first is that it is necessary
to apply heat to the sample prior to each OSL measurement so that all the
measurements are comparable. This preheat, along with the OSL measurement
itself, alters the response of the sample, known as its sensitivity. These changes
Fig. 5. The palaeodose obtained from measurements of many small aliquots,
with between 60 and 100 grains on each, from a range of samples from
south-eastern Australia (from Olley et al. 1998). For fluvial samples where the
degree of bleaching at deposition is lower, the range of palaeodoses is wider
than that from the aeolian sample.
Luminescence dating of sediments using individual mineral grains
93
in sensitivity prevented previous workers from using such an elegant and simple
solution (e.g. Duller 1991; Stokes 1994). Murray & Wintle (2000) overcame
sensitivity changes by developing a method which explicitly monitors the sensi-
tivity during a set of measurements (Fig. 2). After measurement of the OSL
signal relating to the natural dose or one of the laboratory regeneration doses (Lx),
an extra set of measurements are inserted. The aliquot is given a small radiation
dose, called the test dose, heated to 160˚C to remove any signals that would
interfere with the main OSL measurement, and its OSL signal measured (Tx). The
same test dose is used throughout a set of measurements on an aliquot. If no
sensitivity changes occurred then all values of Tx would be identical. In practice
this is not seen, and instead of plotting the raw OSL signal (Lx) to construct
a growth curve, the results are normalised by the response to the test dose (Lx/Tx)
to correct for sensitivity change.
Work by Murray et al. (1995) using an early version of the SAR procedure,
and small aliquots of only a few hundred grains, showed how modern samples
collected from a variety of fluvial depositional environments contained differing
proportions of unbleached, or incompletely bleached, grains (Fig. 3), with an
overbank deposit being the most well bleached. In this case the distribution of
palaeodoses observed from these grains are presented as histograms since it is
known that different quartz grains have very different sensitivities to dose.
Where a mixture of grains is present, the probability of obtaining a sub-sam-
ple containing only well bleached grains decreases rapidly as the number of
grains in the sub-sample increases (Fig. 4, from Olley et al. 1999). Thus in his
early work Li (1994) reduced the number of grains present in his aliquots in order
to accentuate the differences in palaeodose and this approach has also been used
successfully by Olley et al. (1998). Figure 5 shows the distribution of palaeodose
values in aliquots of quartz which contain between 60 and 100 grains (as opposed
to approximately 5000–10000 grains, which is more typical of traditional measu-
rements). These results show the difference in the luminescence signal from
young sediments in fluvial and aeolian depositional environments. A significant
number of the aliquots from the fluvial sediment contain grains whose lumine-
scence signal was not reset at deposition and so give palaeodoses that are much
larger than would be expected for a sample that was less than 5 years old.
Olley et al. (1998) suggested that a distribution of palaeodose values would
be observed where a proportion of the grains within a deposit were not fully reset
at deposition. In this situation, they showed that the best estimate for the palaeo-
dose that has accrued since the last bleaching event was obtained by taking the
average value from the 5% of the aliquots with the lowest palaeodose. This was
tested by applying the method to samples extracted from a sediment core from
the Namoi river. The results were encouraging, though there was no definite age
control.
Lepper et al. (in press) also obtained palaeodose distributions for samples
from different depositional environments using quartz single aliquot measure-
G. A. T. DULLER & A. S. MURRAY
94
ments. They used a deconvolution method to remove any measurement uncerta-
inties and then calculated the palaeodose relating to the latest depositional event
by taking the mean of the value between the lowest palaeodose measurement and
the peak of the distribution. Although this data gave stratigraphically consistent
results, there was no good age control with which to compare the results.
Results from individual grains
The methods used by Olley et al. (1998) and Lepper et al. (in press)
both attempt to separate from a mixed population of mineral grains the palaeo-
dose of those grains which have been bleached during the most recent depositio-
nal event, and so remove the effect of those grains which were not bleached.
A more direct method of achieving this same objective is to make measurements
of the palaeodose from individual grains.
Lamothe et al. (1994) presented palaeodoses for 15 grains of potassium
feldspar extracted from a shallow marine late-glacial sediment in Quebec. In the
case of potassium rich feldspars a significant proportion of the total dose to the
sample originates from the decay of 40K within the grain. Particularly large grains
(750–1000 µm diameter) were used in this study in order to facilitate manipula-
tion of the grains by hand during luminescence measurement. As a consequence,
at least 50% of the total dose rate arose from within the grains. The ages calcu-
lated for the 15 grains varied from 700%, to approximately 70% of the estimated
age. The presence of grains with significant age overestimates demonstrated the
benefit of single grain analysis for inadequately bleached samples. It was more
difficult to explain the presence of grains which underestimated the expected age,
but this may have been caused by a phenomenon termed ’anomalous fading’
which can affect measurements of feldspars (e.g. Spooner 1994).
Murray & Roberts (1997) were the first to obtain palaeodoses from individual
sand-sized grains of quartz. In their samples from Australia they noted that
within a single sample they observed a broad range of OSL signal intensities,
varying by several orders of magnitude. The range of palaeodose values from
a single sample was also broad, but for these sub-aerial samples the values were
consistent with a single age for the sediment.
More recently, the benefit of single grain analysis has been demonstrated very
clearly by the work of Roberts et al. (1999) at a site called Jinmium, in north-
west Australia. Jinmium is an important Aboriginal rock shelter site, with ar-
chaeological evidence of human activity. The site was first brought to prominen-
ce by Fullager et al. (1996) where thermoluminescence dates (using large aliqu-
ots of ~10 000 grains) measured on sands collected from an excavation at the
base of the rock shelter suggested that humans had arrived at this site prior to
116±12 ka, almost 50 kyr earlier than previous estimates for the first human
arrival in the continent (Roberts et al. 1994). However, the site at Jinmium is
Luminescence dating of sediments using individual mineral grains
95
a complex one for luminescence dating. The ‘rock shelter’ consists of a slightly
overhanging sandstone rock face. Blocks of sandstone that had fallen from the
overhanging face were encountered during the excavations at the base, mixed
together with sand that was thought to have blown into the site from the surro-
unding area. If the blocks disintegrated in situ then the grains released would not
be exposed to daylight. The presence of such grains mixed with those that were
delivered by aeolian processes would lead to an overestimate of the age. Single
grain analysis suggested that this had occurred (Roberts et al. 1998) and that if
only those grains from the younger population were included in the analysis then
a depositional age of younger than 10 ka was more appropriate.
In the last year, single grain analysis has also been used for dating fluvial
sediments. Exposure of mineral grains to sunlight during transport may occur on
a discrete basis, but in certain fluvial settings some grains may remain unble-
ached. As with Jinmium, the application of standard multiple grain analyses will
result in an age overestimate. Olley et al. (1999) have shown that such heteroge-
neous bleaching does occur in some fluvial deposits, and that single grain analy-
sis provides a means to obtain accurate depositional ages.
Practical difficulties
Analysis of single mineral grains 100–300 µm in diameter is complex.
The luminescence signal from single grains tends to be weak, and manipulating
such grains in laboratory conditions is difficult (all samples have to be handled
under dim red light prior to analysis in order to prevent loss of the light-sensitive
luminescence signal). Three methods have previously been developed to cope
with the analysis of single grains. The first, and most widely used, is to hand-pick
individual grains and mount them on standard aluminium sample holders. These
holders are 9.7 mm in diameter and are designed to accommodate many thousand
grains. However Lamothe et al. (1994) and Murray & Roberts (1997) have
successfully used this procedure for single grains. The advantage is that conven-
tional luminescence equipment can be used for the measurements. However, the
drawbacks are that it is laborious and that it makes very heavy use of instrument
time.
A second approach has been to use an imaging system, such as an imaging
photon detector (e. g. McFee 1998a) or a charge coupled device camera (e.g.
Duller et al. 1997). Using such systems it is possible to mount many grains onto
a single aluminium sample holder and then produce a two dimensional ’image’
of the luminescence signal. Thus one can resolve the luminescence signal coming
from individual grains. This has the advantage over hand picking grains that
many grains can be analysed simultaneously. However these measurement sy-
stems are technically complex and expensive. Additionally, a critical issue is
how reliably repeated measurements can be made on a single grain. This is
G. A. T. DULLER & A. S. MURRAY
96
Fig. 6. Diagram of the optical stimulation section of the single grain
luminescence system constructed at Risø. The path of the laser beam is shown
by the dotted lines. It is focussed using a series of three lenses, and its precise
position on the sample is controlled by moving the two mirrors. Single
mineral grains are held in a nine-by-nine array of holes drilled into the surface
of an aluminium disc. Further details of how the system works are given in
Duller et al. (1999a, b).
Luminescence dating of sediments using individual mineral grains
97
essential for measurement of the palaeodose from a grain, yet McFee (1998b)
calculated that the system based around an IPD had a 25% measurement uncer-
tainty. However, in spite of these problems, such systems have been employed
to measure palaeodoses (McFee 1998a).
A third approach, described by Bailiff et al. (1996), involved moving a stan-
dard aluminium sample holder under a focused laser beam so that each point on
the sample was illuminated by the laser spot in turn, and the resulting OSL signal
measured. While promising initial results were presented, the measurement time
was prohibitive; the total scanning time for a single sample holder was at least
one hour, and many such scans are required to derive palaeodoses from a single
holder.
New technology
In the last two years a new piece of equipment specifically designed
to make single grain luminescence measurements has been designed and built
(Duller et al. 1999a, b). The system has two key features. The first is that the
Fig. 7. An optically stimulated luminescence decay curve from a single grain
of quartz extracted from a Tasmanian dune sand (TNE9503). A series of
additional OSL decay curves were measured from the same grain after various
treatments, following the procedure outlined in Fig. 2. The data were used to
calculate a single aliquot regeneration (SAR) growth curve. This is shown as
an inset to the main diagram. The corrected natural OSL signal (Lx/Tx) is
shown as a solid square, while the regenerated signals used to define the
growth curve are shown as solid circles.
G. A. T. DULLER & A. S. MURRAY
98
standard 9.7 mm aluminium sample holder has been modified so that it contains
an array of 81 holes drilled into its surface (Fig. 6). Each hole is 300 µm wide
and 300 µm deep. Each hole centre is accurately drilled so that the grains lie 600
µm apart. The second key feature is a shuttered laser beam (532 nm, 10 mW)
that is focused to a 20 µm diameter spot at the sample holder. This beam enters
the measurement chamber via two mirrors which can be moved under computer
control, and so allow the beam to be steered to any position on the sample holder.
Thus the system can direct the beam at any one of the eighty-one grains mounted
on the sample holder and stimulate OSL from that grain. This signal is then
detected using a standard photomultiplier tube. The new single grain system has
been designed to attach to an existing automated Risø TL/OSL reader so that it
can benefit from having an automated sample changer, a heater stage that allows
thermal pretreatment of the sample, and a beta source for irradiation. The total
system can perform all the measurements necessary for palaeodose calculations
under computer control without the need for an operator. Since the automatic
sample changer can cope with up to 48 samples, and each sample holder can
accommodate 81 grains, the system has a maximum capacity of just under 4000
grains, with a typical OSL measurement time of only about 200 s per disc.
Figure 7 shows an OSL decay curve measured using this automated system.
This is the OSL signal from a single 180–211 µm diameter grain of quartz. The
inset to the figure shows how the data from several such OSL measurements on
the same grain can be used to construct a growth curve and hence to calculate
the palaeodose.
Examples of single grain analysis
Grain brightness
Utilising the new equipment described above, it is possible to make
luminescence measurements of many tens or hundreds of single grains. Duller et
al. (in press) have used this instrument to measure the OSL signal from many
grains within a sample. The general feature of all of these measurements is that
there is a very large variability in the intensity of the luminescence signal obtai-
ned from different grains of the same sample. Previous authors have presented
similar data as histograms of grain brightness (e.g. McFee & Tite 1998), but this
is now considered inadequate because of the large dynamic range observed. An
alternative way to present the data is as a cumulative sum, ranking the grains in
order of descending brightness, and then plotting the cumulative light sum as
a function of the proportion of the brightest grains involved (Fig. 8). If all grains
gave the same OSL signal then a diagonal line would be plotted from the origin
to the upper right hand corner of the diagram. In practice all samples will fall
Luminescence dating of sediments using individual mineral grains
99
above and to the left of the line. The further that a sample plots away from the
‘ideal’ diagonal line, the less even the distribution of signal within the different
grains making up the sample. For instance, in samples BA14 and RBM2 (coastal
aeolian sands from southern Africa), over 95% of the OSL signal originates from
less than 5% of the grains, and the majority of the grains play a minor role in the
overall signal. In contrast, WIDG8 (an aeolian sand from northern Australia)
contains many grains that contribute.
Figure 4 showed that where a mixture of well-bleached and unbleached gra-
ins existed, the probability of selecting only well-bleached grains decreased
rapidly as the number of grains increased. For multiple grain work this will give
scatter in the palaeodose values obtained. If one compared the multiple grain
Fig. 8. The proportion of the total OSL light sum from a set of grains plotted
as a function of the proportion of the brightest grains that are used. For
a population of grains which all have the same brightness, the line would run
diagonally from bottom left to top right. For all natural samples the line plots
to the left of this. Data from a range of samples are shown. The number of
grains measured from each sample is shown in brackets. Details of the
samples are given in Duller et al. (in press).
G. A. T. DULLER & A. S. MURRAY
100
behaviour of two samples, such as BA14 with a few very bright grains and many
dimmer ones, and WIDG8 with many similarly bright grains, they would be very
different since the effective number of grains on an aliquot of equal mass is very
different. This is an additional reason why single grain analysis is preferable to
multiple grain analysis where there is a mixture of well-bleached and unbleached
grains.
Palaeodose distributions
Figure 9 shows a set of palaeodoses obtained from 408 grains of
a dune sand from north-east Tasmania. The sample dates from the last glacial
maximum, and has not undergone any post-depositional reworking. An important
consideration when analysing single grain data is how to present the results. As
shown above, grains of quartz from a single sample may have OSL signals which
vary by over two orders of magnitude. As well as affecting the multiple grain
behaviour, this will influence the precision with which the palaeodose can be
calculated. For the brighter grains the uncertainty may be 5%, while for dimmer
grains it might be 100%. For such data it is unreasonable to present it as a histo-
gram since this implicitly assumes that each data point should be weighted
equally. Two alternative methods of presentation are possible. The first is to
produce a probability density function (PDF). This is a weighted histogram whe-
re each palaeodose is represented by a gaussian curve whose peak is at the
palaeodose value, but whose height is inversely related to the precision with
which the palaeodose is known. Hence the result from a grain whose palaeodose
is well known is represented by a narrow, high peak, while a grain whose palaeo-
dose is poorly known is represented by a low broad peak. The results for all
grains are then summed to produce a single probability density function (Fig. 9a).
This approach has been criticised because it makes it impossible to discern the
influence of an individual data point upon the overall distribution. Instead, Gal-
braith (1990) suggested using a radial plot (Fig. 9b). On this graph, each data
point is represented discretely. Full details of how such plots are constructed are
given by Galbraith (1990). The essential points are that the greater the precision
with which a palaeodose is known, the further it is plotted to the right of the
graph. The difference between some average palaeodose value and the value for
the specific grain, divided by the standard error on that specific palaeodose
dictates the vertical position of the point on the graph. A consequence of the
quantities that are plotted is that any points lying on a line drawn through the zero
point on the y-axis all have the same palaeodose. Thus a radial scale can be
added on the right hand side of the graph, marked with values of palaeodose.
An additional advantage of the radial plot is that one can draw two parallel
lines from the values +2σ and –2σ on the y-axis to intersect the radial axis on
the right hand side. Any data point whose palaeodose is consistent, within two
Luminescence dating of sediments using individual mineral grains
101
Fig. 9. Palaeodose measurements from 408 grains of quartz extracted from
a Tasmanian dune sand (TNE9503). The palaeodose values are plotted (a) as
a probability density function and (b) as a radial plot. The way in which such
plots are constructed is described in the main text.
G. A. T. DULLER & A. S. MURRAY
102
Fig. 10. Palaeodose measurements from 37 grains of quartz extracted from
a frost-wedge cast, Denmark (992101). The palaeodose values are plotted as
(a) a probability density function and (b) a radial plot.
Luminescence dating of sediments using individual mineral grains
103
standard deviations, with the mean palaeodose value used to construct the radial
plot will fall within the band defined by these two lines. In the case of the aeolian
sand from Tasmania, 81% of the grains do indeed fall within this band, demon-
strating that the grains from the sample do yield a set of palaeodose values which
are consistent with a single value of 22.8±0.4 Gy, implying that the sample was
well bleached at deposition.
In contrast, analysis of single grains from a sand infilling a frost-wedge cast
(sample 992101) give a wide range of palaeodose values, and the radial plot
(Fig. 10b) shows that only 51% of these are consistent with the mean value
within two standard deviations. This sample clearly consists of at least two
populations with a peak in the probability density function at approximately 50
Gy (Fig. 10a).
Conclusions
Luminescence dating methods have been applied successfully to sedi-
ments from a wide range of Quaternary sites over the last two decades. The
development of single aliquot procedures for palaeodose determination has in-
creased the throughput of samples, and reduced the analytical uncertainties.
The most successful depositional environments for luminescence dating are,
not suprisingly, those in which the probability of grains being exposed to daylight
at deposition is large, such as coastal and desert dunes (Wintle 1993) and loess.
In such environments it is reasonable to assume that all grains have been equally
bleached. However, there are a wide range of environments in which it is possib-
le that grains have been adequately exposed to daylight, but it would be unwise
to assume this to be the case. Sediments deposited by fluvial, glacio-fluvial and
mass movement processes are likely to consist of a complex mixture of grains,
with only a proportion having been completely exposed. In such situations the
luminescence age calculated using standard multiple grain procedures will be an
overestimate. A variety of methods have been developed which will identify
such complex situations from single aliquot data, but which are unable to provide
an indication of the extent of the overestimate. In such situations aliquots with
a restricted number of grains may provide an estimate of the true palaeodose.
A more direct method is to analyse single grains in order to directly differentiate
between those grains which have been bleached and those which have not.
The development of single aliquot and single grain analytical procedures
provide a new avenue of research in luminescence dating. The ability to analyse
the luminescence properties of individual mineral grains within a sample widens
the range of depositional environments in which the method can be applied, and
provides far greater information about the sample than was available previously.
Routine analysis of single grains is also being made practical by the development
of instrumentation specifically designed for this purpose.
G. A. T. DULLER & A. S. MURRAY
104
Acknowledgements: This work is funded by the Danish Natural Science Research Council
(9701837) and NERC (GR3/E0087). Dr Helen Roberts suggested valuable improvements to an
early version of the manuscript. The support and encouragement of Lars Bøtter-Jensen and Ann
Wintle throughout our pursuit of single grain measurements has been greatly valued.
References
AITKEN M. J., 1998: An introduction to optical dating. Oxford University Press.
BAILIFF I. K., CLARK R. J. & BARNETT S. M., 1996: A high resolution scanning system
for measuring spatially resolved optically stimulated luminescence. Abstract: 8th
International Conference on Luminescence and Electron Spin Resonance Dating,
Canberra, Australia.
DULLER G. A. T., 1991: Equivalent dose determination using single aliquots. Nuclear
Tracks and Radiation Measurements, 18: 371–378.
– 1994a: Luminescence dating using single aliquots: new procedures. Quaternary Geo-
chronology (QSR), 13(2): 149–156.
– 1994b: Luminescence dating of poorly bleached sediments from Scotland. Quaterna-
ry Geochronology (QSR), 13: 521–524.
– 1996: Recent developments in luminescence dating of Quaternary sediments. Pro-
gress in Physical Geography, 20: 127–145.
DULLER G. A. T., BØTTER-JENSEN L. & MURRAY A. S. (in press). Optical dating of
single sand-sized grains of quartz: sources of variability. Radiation Measurements.
DULLER G. A. T., WINTLE A. G. & HALL A. M., 1995: Luminescence dating and its
application to key pre-Late Devensian sites in Scotland. Quaternary Science Reviews,
14: 495–519.
DULLER G. A. T., BØTTER-JENSEN L., KOHSIEK P. & MURRAY A. S., 1999a: A high-
sensitivity optically stimulated luminescence scanning system for measurement of
single sand-sized grains. Radiation Protection Dosimetry, 84: 325–330.
DULLER G. A. T., BØTTER-JENSEN L., MURRAY A. S. & TRUSCOTT A. J., 1999b: Single
grain laser luminescence (SGLL) measurements using a novel automated reader.
Nuclear Instruments and Methods: B 155: 506–514.
DULLER G. A. T., BØTTER-JENSEN L. & MARKEY B. G., 1997: A luminescence imaging
system based on a charge coupled device (CCD) camera. Radiation Measurements,
27: 91–99.
FULLAGER R. L. K., PRICE D. M. & HEAD L. M., 1996: Early human occupation of
northern Australia: archaeology and thermoluminescence dating of Jinmium rock-
shelter, Northern Territory. Antiquity, 90.
GALBRAITH R. F., 1990: The radial plot: graphical assessment of spread in ages. Nuclear
Tracks and Radiation Measurements, 17: 207–214.
HUNTLEY D. J. & BERGER G. W., 1995: Scatter in luminescence data for optical dating
– some models. Ancient TL, 13: 5–9.
LAMOTHE M., BALESCU S. & AUCLAIR M., 1994: Natural IRSL intensities and apparent
luminescence ages of single feldspar grains extracted from partially bleached sedi-
ments. Radiation Measurements, 23: 555–561.
LEPPER K., AGERSNAP LARSEN N. & MCKEEVER S. W. S. (in press), Application of
Luminescence dating of sediments using individual mineral grains
105
equivalent dose distribution analysis to Holocene eolian and fluvial quartz sands from
Central Oklahoma. Quaternary Geochronolgy (QSR).
LI S.-H., 1994: Optical dating: insufficiently bleached sediments. Radiation Measure-
ments, 23: 563–567.
MCFEE C. J., 1998a: The measurement of single grain IRSL EDs using an imaging
photon detector. Quaternary Geochronology (QSR), 17: 1001–1008.
– 1998b: Some methodological techniques for luminescence imaging using an IPD.
Ancient TL, 16: 11–18.
MCFEE C. J. & TITE M. S., 1998: Luminescence dating of sediments – the detection of high
equivalent dose grains using an imaging photon detector. Archaeometry, 40, 153–168.
MURRAY A. S., 1996: Developments in optically stimulated luminescence and photo-
transferred thermoluminescence dating of young sediments: application to a 2000
year sequence of flood deposits. Geochemica et Cosmochimicha Acta, 60: 565–576.
MURRAY A. S., OLLEY J. M. & CAITCHEON G. G., 1995: Measurement of equivalent
doses in quartz from contemporary water-lain sediments using optically stimulated
luminescence. Quaternary Geochronology (QSR), 14: 365–371.
MURRAY A. S. & ROBERTS R. G., 1997: Determining the burial time of single grains of
quartz using optically stimulated luminescence. Earth and Planetary Science Letters,
152: 163–180.
MURRAY A. S., ROBERTS R. G. & WINTLE A. G., 1997: Equivalent dose measurement
using a single aliquot of quartz. Radiation Measurements, 27: 171–184.
MURRAY A. S. & WINTLE A. G. (2000) Luminescence dating of quartz using an impro-
ved single-aliquot regenerative-dose protocol. Radiation Measurements, 32: 57–73.
OLLEY J. M., CAITCHEON G. G. & MURRAY A. S., 1998: The distribution of apparent
dose as determined by optically stimulated luminescence in small aliquots of fluvial
quartz: Implications for dating young sediments. Quaternary Geochronology (QSR),
17: 1033–1040.
OLLEY J. M., CAITCHEON G. G. & ROBERTS R. G., 1999: The origin of dose distributions
in fluvial sediments, and the prospect of dating single grains from fluvial deposits using
optically stimulated luminescence. Radiation Measurements, 30: 207–217.
ROBERTS R. G., BIRD M., OLLEY J., GALBRAITH R., LAWSON E., LASLETT G., YOSHIDA H.,
JONES R., FULLAGER R., JACOBSEN G. & HUA Q., 1998: Optical and radiocarbon
dating at Jinmium rock shelter in northern Australia. Nature, 393: 358–362.
ROBERTS R. G., JONES R., SPOONER N. A., HEAD M. J., MURRAY A. S.& SMITH M. A.,
1994: The human colonisation of Australia: Optical dates of 53,000 and 60,000 years
bracket human arrival at Deaf Adder gorge, Northern Territory. Quaternary Geo-
chronology (QSR), 13: 575–583.
SPOONER N. A., 1994: The anomalous fading of infrared-stimulated luminescence from
feldspars. Radiation Measurements, 23: 625–632.
STOKES S. 1994: The timing of OSL sensitivity changes in a natural quartz. Radiation
Measurements, 23: 601–605.
WINTLE A. G., 1993: Luminescence dating of aeolian sands – an overview. [In:] The
dynamics and environmental context of aeolian sedimentary systems. (K. Pye, Ed.),
pp. 49–58. Geological Society Special Publication, London.
WINTLE A. G., LI S.-H. & BOTHA G. A., 1993: Luminescence dating of colluvial depo-
sits from Natal, South Africa. South African Journal of Science, 89: 77–82.
G. A. T. DULLER & A. S. MURRAY
106
